arjunkshah12345-hash

Agent Orchestrator Bench — Testing skill for Claude Code

Testing community

Hands-on benchmark of agent orchestrators for coding agents (docs + feature testing).

How to install Agent Orchestrator Bench

This entry records only its repository, not the path inside it, so there is no exact command to give. Open arjunkshah12345-hash/agent-orchestrator-bench and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Orchestrator Bench does

Hands-on benchmark of agent orchestrators for coding agents (docs + feature testing).

Alternatives in Testing

Spy — by spylang - Enforces strict coding conventions with comprehensive testing guidelines, multiple code compilati 678 ★
Production Grade — Production Grade — route any build, feature, review, test, harden, ship, or docs request through the 14-agent 175 ★
Robotics Agent Skills — Agent skills that make AI coding assistants write production-grade robotics software 170 ★

README

Agent orchestrator bench

A feature and orchestration benchmark (not a model leaderboard) of the apps that run several coding agents (Claude Code, Codex, Cursor, OpenCode and others) in parallel with worktrees, sessions and review. 22 tools were checked, 13 scored on a 9-part rubric, 2 driven end-to-end. Snapshot date: 2026-10-19.

**Live site:** https://arjunkshah12345-hash.github.io/agent-orchestrator-bench/

Measured: stub-agent test on Linux (3 runs each)

There's no universal winner. Orca is faster to launch; Claude Squad is far lighter.

	Orca	Claude Squad
Verified stub tasks	18/18	18/18
Six-agent launch (s)	1.52 to 1.90	11.18 to 11.89 (includes about 1.1 s per session of our tmux driver delays)
Peak RSS (MB)	1124 to 1140 (about 1.1 GB)	102 to 118

The stub agent is a deterministic script, not an LLM. It measures the orchestrator, not code quality. Logs: `runs/repeat/`.

Secondary research: weighted feature assessment

This is a weighted feature assessment from docs, repos and feature testing, not a measured performance leaderboard. Orca was run end-to-end with stub agents; T3 Code was not run through an equivalent session, so 93 vs 90 is not experimentally proven. Criteria without evidence are unranked; the range shows the score with them at 0 or at full marks, and the point score is the midpoint.

#	Tool	Assessed /100	Range	Doc coverage	Hands-on validation
1	Orca	93.0	93.0 to 93.0	100%	Stub end-to-end
2	T3 Code	90.0	90.0 to 90.0	100%	Launched
3	Synara	79.0	79.0 to 79.0	100%	Launched
4	Emdash	74.5	72.0 to 77.0	95%	Launched
5	Paseo	74.5	57.0 to 92.0	65%	Launched
6	OpenChamber	74.0	64.0 to 84.0	80%	Launched
7	Superset	72.5	70.0 to 75.0	95%	Launched
8	Cloudroom (David Ondrej)	72.0	62.0 to 82.0	80%	Launched
9	Vibe Kanban	69.0	69.0 to 69.0	100%	Launch